Enzymes That Destroy Blood Group Specificity

Evidence is presented for the presence of an enzyme, a-N-acetylgalactosaminyl oligosaccharase in filtrates of Clostridium perfringens. Action of this enzyme on hog submaxillary glycoprotein results in the release of a disaccharide, characterized as 3-0-fi-n-galactopyranosyl-2-acetamido-2-deoxy-o-galactose. The possibility that this oligosaccharide has resulted from tranagalactosylation has been eliminated. Enzymatic treatment of several glycoproteins and blood group substances from miscellaneous sources also resulted in the production of this oligosaccharide. Since in the blood group substances N-acetylgalactosamine at the reducing end of the oligosaccharide chain is o-linked to the hydroxy amino acids of the protein core, ue have here a very useful cr-N-acetyl-D-galactosaminyl oligosaccharase for structural studies of the carbohydrate-protein linkage in the blood group substances and in glycoproteins.

The possibility that this oligosaccharide has resulted from tranagalactosylation has been eliminated.
Enzymatic treatment of several glycoproteins and blood group substances from miscellaneous sources also resulted in the production of this oligosaccharide. Since in the blood group substances N-acetylgalactosamine at the reducing end of the oligosaccharide chain is o-linked to the hydroxy amino acids of the protein core, ue have here a very useful cr-N-acetyl-D-galactosaminyl oligosaccharase for structural studies of the carbohydrate-protein linkage in the blood group substances and in glycoproteins.
In our investigations on the action of a crude preparation of cu.N-acetyl-n-galactosamillidase from Clostridium perjringens on hog submaxillary glycoprotein with blood group A specificity (I), we noted the anticipated release of free N-acetylhesosamirre. The amount of the acyl amino sugar released was determined by a modification (2) of the YIorgan-Elson reaction (3). Prolonged incubatioii of the enzyme with substrate resulted in the complete release of the N-acet~ylgalt\c:touarrline.
However, the amount of color obtained in the calorimetric test, was far in excess of t.he total amount of hr-acetylgalactosarnille present in the molecule, as determined by the hexosamine content after complete acid hydrolysis (4). Indeed, the color obtained was even more than could be accounted for by the simultaneous release of N-glycolylnlannosamine from the sialic acid, as a consequence of the action of sialidase and N-acetylneuraminic acid-aldolase. Both of these enzymes are known to be present in crude extracts of C. perfringens (5,6). * This work was supported try Grant AM-07305 from the National Institutes of Health and in part, by a grant from the Horace Ii. Rackham School of Graduate Studies, t,he University of Michigan. The fourt,h paper in this series, "Effect of Sinlidase on Blood Group Specificity of IIog Submaxillary (:lyc:oprotein," is Reference 9. Similar results were obtained when the A-H--and H-specific submaxillary glycoproteins from hogs were used (1). On the other hand, prolonged incubation of the enzyme preparation with ovine submaxillary glycoprotein, or the neutral tetrasaccharide, 2-acetamido-cu-n-galactopyranosyl- (1 + 3).[cu-n-fucopyranosyl-(1 ---) 2)]-/3+galactopyranosyl-(1 -+ 3)-2-acetamido-2-deoxy-ngalactitol, isolated from the alkaline borohydride degradation products (7) of the hog A-specific submaxillary glycoprotein, gave almost the theoretically expected amount of N-acetylgalactosamine.
Studies to determine the nature of the "excess" Morgan-Elson positive-reacting material resulted in the identification in C. perjringens of an oligosaccharase t.hat is capable of hydrolyzing the linkage of oligosaccharide chains to the polypeptide backbone of the glycoprotein. This report will present our data in support of the belief that the enzyme involved is an oligosaccharase, and not a galactosyltransferase, and that the reaction product formed is 3-~-~-r~-gaIactopyranosyl-2-acetamido-2-deosy-n-galactose.
N-Acetylgalactosaminitol and galactosaminitol were prepared according to the procedure of Crimmin (8).
A commercial preparation of hyaluronidase (Worthington Biochemical Co.) (14)4 was used as the source of P-r-galactosidase. D-Galactose dehydrogenase was purchased from Boehringer and Soehne, Mannheim, Germany.
The crude preparation of enzymes from C. perfTingens was obtained as described elsewhere (15). This preparation was further fractionated by the use of ammonium sulfate at concentrations of between 50 and 520/u saturation which was essentially free of N-acetylneuraminic acidaldolase (6), but still contained sialidase (5) and fucosidase (15).

Methods
Galactose was determined by the anthrone rnethod (16), in which both galactose and fucose have the same molecular extinction coefficients, and corrected for the total fucose content of the specimen.
Total fucose was determined after a lo-min heating period (17). Total sialic acid was assayed by a modification (5) of the Svennerholm procedure (18). The nitrogen contents of the oligosaccharides were determined by a ninhydrin method after digestion with sulfuric acid (19).
Oligosaccharides or glycoproteins were hydrolyzed with 0.5 N HCl for 4 hours at 100" in sealed ampules (4) for qualitative chromatographic and electrophoretic analyses as well as for quantitative analyses.
Chromatography and electrophoresis of the oligosaccharides and sugars were performed on Schleicher and Schuell SS No. 589 green ribbon paper. Solvents used for irrigation in the paper chromatography were: Solvent 1, l-butanolpyridine-water (6:4 :3 v/v); Solvent 2, phenol saturated with water.
Paper electrophoresis was performed in 1% sodium borate buffer, pH 9. The oligosaccharides and free sugars were detected by the use of AgNOa-NaOH, while the hexosamines and N-acetylhexosamines were detected with Ehrlich reagents (21). The total hexosamine content was determined by the Rondle and Morgan modification (22) of the Elson-Morgan (23) reaction on a neutralized hydrolysate of the glycoprotein or oligosaccharide.
N-Acetylhexosamines were determined by the Mogan-Elson reaction (3) using the modification of Hansen (2) adapted to a micro scale. Under these conditions for the determination, N-acetylglucosamine gives 224%, N-acetylmannosamine 123%, and N-glycolylmannosamine 139% of the color given by MI equimolar amount of N-acetylgalactosamine.
Structural studies involved determination of the amount of periodate reduced (24) at room temperature over a period of time at pH 4.5, 0.1 M sodium acetate, in the dark.
The amount of formaldehyde released was determined by a chromotropic acid procedure (25) and the amount of formic acid by the thiobarbituric acid method (26); the determinations were carried out simultaneously with the studies on periodate consumption.
Values reported represent the maximal amounts.
The amount of color produced is expressed as a percentage of that theoretically expected on the basis of t.he N-acetylgalactosamine content of the material. After 24 hours of incubation at 37", both hog A+-and H+-specific submasillar\ glycoproteins released "excess" ME+ material (1800/;, and 205y0, respectively), whereas ovine submaxillary glycoprotein released 126% and the neutral tetrasaccharide from hog :I+-specific glycoprotcin, only 88.5%. III order to determine the nature of the components giving this "excess" color in the Morgan-Elson reaction, 24-hour incubation mixtures were treated with ethanol to a final concentration of 800/, to stop the reaction and to precipitate the salts and denatured proteins.
The supernatant solutions were concentrated and, along with standard sugars, chromatogmphed in duplicate for 6 hours in Solvent 1. The chromatograms were t.hen developed for reducing sugars and ME+ material, respectively (Table I). Both A+-and H+-specific hog submasillary glycoproteins gave rise to three ME+ reacting components which also reacted as reducing sugars with the silver nitrate reagent.
The mobilities of Components A and C corresponded to N-glycolylneuraminic acid and N-acetylgalactosamine. Sheep submaxillary glycoprotein gave only two ME+ spots, corresponding to N-acetylneuraminic acid and N-acetylgalactosamine. Free galactose and fucose were also detected in these enzymatic hydrolysates, when incubation mixtures containing hog submaxillary glycoproteins were used. The unknown ME + reacting material obtained from the hog submaxillary glycoproteins, Component B, aIso reduced alkaline silver nitrate and had an RN (mobility of spot relative to the mobility of N-acetylgalactosamine taken as unity) of 0.52 to 0.59.
Isolation-For further characterization of Component B, 1 g of the At-specific hog submaxillary glycoprotein was enzymati-

6739
tally hydrolyzed under the above conditions in a final voIume of 116 ml for 24 hours at 37". The reaction was stopped by heating in a boiling water bath, and the cooled solution was treated with absolute ethanol to a final concentration of 800/0, cooled to -20" overnight, and centrifuged at 18,000 x g for 15 min. The supernatant was concentrated by vacuum distillation and applied to eight sheets (24 x 58 cm) of SS 589 green paper.
The papers were irrigated with Solvent 1 for 6 hours.
Appropriate guide strips were cut on either side of the sheets and sprayed with Ehrlich reagent to locate the ME+ reacting material.
The unknown component, with an RN of 0.55 to 0.57, was then eluted from the appropriate area of the unstained sheets. The eluat,es from the eight sheets were pooled, concentrated in uucuo to a syrup, and further purified by passage through a column (2.5 x 38 cm) of Sephadex G-15 to remove a faintly yellow contaminant.
The final yield of product was 170 umoles.
Purity-The purity of the component with an RN of 0.56 was established by obt.aining only one spot on chromatography in Solvent I, RN 0.55, and in Solvent 2, RN 0.855, and on electrophoresis in 1 y0 sodium borate buffer, pH 9. In all cases only one spot was obtained with the same RN value with both developing agents, alkaline silver nitrate and Ehrlich reagent for N-acetylhexosamine.
Composition-Quantitative analysis of the purified unknown compound for component sugars indicated the absence of any sialic acid or fucose. Galactose and Morgarl-Elsoll-1)ositive material were present equivalent to 15.3 and 30.6 prnoles per ml, respectively.
The total nitrogen content was 17 Mmoles per ml. Acid hydrolysis follow-cd by paper chromatography revealed the presence of galactose and galactosamine only. Quantitative analysis of the hydrolysis products indicated the presence of 15.3 pmoles per ml of galactose as determined by galactose dehydrogenase and by gas liquid chromatography (27), and 16.0 prnoles per ml of galactosamine as determined by the Elson-Morgan reaction and by the use of an amino acid analyzer (28) ; the total nitrogen content was 16.4 pmoles per ml. On the basis of these analytical data, we can conclude that the unknown compound is a disaccharide of galactose and galactoxamine.
Hydrolysis of the oligosaccharide with P-galactosidase from bovine testes and with the P-galactosidase in commercial hyaluronidase preparations resulted in the release of galactose, as determined by galactose dchydrogenase, and in the release of N-acetylgalactosamine.
The presence of these two component sugars was further established by chromatography; both the RF values and reactivity wit,h alkaline silver nitrate and Ehrlich reageilts identified the sugars with the standards run in parallel.

Strucfure of Oligosaccharides
Reduction of Oligosaccharide with NaBI14~~'l%c oligosaccharide (16 ~molcs) was reduced with excess sodium borohydride (2700 pmoles) ill :I final volume of 2 ml iit au ice bath For 18 hours and in the dark (25). The excess borohydride was then destroyed by titration with 4 PJ HCl to pH 6. The salt was removed by passage through a column (2.5 X 38 cm) of Dowcx 50 (H+, 200 to 400 mesh) and eluted with 500 ml of water.
The eluatc was cvaporat,od to dryness and the boric acid removed as its methyl ester (29).
Oxidation of Oligosurcharide with NuO-The oligosaccharide (23 ~rnoles) ww oxidized with NaOI ill an iilc:ul,a&ll mixture containing 2.5 7711 of 0.1 N I, iu 0.2 M KI and 0.9 ml of 50/O N&03 in a final volume of 10 ml, according to the analytical procedure of Jlaclcod and Rob&n (30). The mixture was set aside in t.he dark at room temperature for 2 hours to completely oxidize the oligosaccharide.
The solution was then acidified with 4 N acetic acid and the resulting I1 was successively extracted with several aliquots of chloroform.
The aqueous layer was then concentrated in YUCUO and passed through a column (2.5 x 38 cm) of Sephadex G-15 to remove low molecular weight contaminants.
Composition of Reduced and Oxidized Oligosaccharides-The ability of the original oligosaccharide to give a positive Morgan-Elson reaction was completely abolished in the reduced and oxidized derivatives respectively, whereas the reactivity of galactose in the anthrone reaction was unimpaired.
Hydrolysis of the reduced and oxidized oligosaccharides followed by chromatography revealed the presence of free galactose as determined by RF and, in each case, a second compound that reacted with the alkaline silver nitrate but gave no reaction with the Ehrlich reagents, and corresponded to galactosaminitol and galactosaminic acid, respectively.
Oxidation with Period&e-Oxidation of 1 pmole of oligosaccharide with excess periodate for 24 hours resulted in the consumption of 4.6 Imoles of periodate and the release of 1.08 prnoles of formaldehyde and 2.05 pmoles of formic acid.

Tests for Possible Transglycosylation
The disaccharide could arise as a consequence of transglycosylation since both galactose and N-acetylgalactosamine were detected in a 24-hour incubation mixture.
The following experiments were therefore set up to eliminate that possibility.
1. Incubation of Free Sugars with Enzymes from 6. perfringens-An artificial mixture of sugars was prepared.
The proportions of the sugars in this mixture were as present in the A+-active hog submaxillary glycoprotein (4.4 pmoles of galactose, 2.2 pmoles of fucose, 4.1 pmoles of N-glucolylneuraminic acid, and 7.75 pmoles of N-acetylgalactosamine).
The rnixture was incubated with the C. perjringens preparation under the conditions described for the preparation of the oligosaccharide.
A similar incubation mixture was set up containing the same sugars and in the same proportions, with the esception of the replacement of N-acetylgalactosamine by /3-p-nitrophenyl-N-acetylgalactosamitiide.
The two incubation mixtures were maintained at 37" for 24 hours. Aliquots taken at the beginning and end of the incubation period showed no change in the intensity of color obtained in the Morgan-Elson test. The reactions were stopped by the addition of alcohol.
Chromatographic analysis of the 80% ethanol supernatants of both incubation mixtures in both solvents, 1 and 2, likewise gave no evidence of the presence of the oligosaccharide.
The incubation products were chromatographed as in Section 1 above for the detection of disaccharide.
No galactosy-N-acetylgalactosaminide disaccharide was detected with lactose or melibiose as galactosyl donor even though both these oligosaccharides were hydrolyzed by the crude enzyrne preparation. Some galnctosyl-N-acctylgalactosaminide was detected with the oligosaccharide alditol mixture (d) but the intensity was less than that obtained with the intact A+ glycoprotein (c) (cf . Table III). hloreover, the amount of disaccharide formed was about the same 6740 when the A+ oligosaccharide alditols were incubated alone with the enzyme and without the addition of excess N-acetylgalactosamine or P-p-nitrophenyl-N-acetylgalactosaminide.
These results would suggest an incomplete alkaline borohydride degradative cleavage of the A+ glycoprotein, rather than the possible synthesis of disaccharide by transglycosylation.
In order, therefore, to further substantiate that there is no net synthesis of the galactosyl-N-acetylgalactosaminide by transglycosylation, the following experiments were carried out using radioactively labeled galactose and N-acetylgalactosamine.
3. Incorporation Studies with Radioactive Galactose--Incubation mixtures of A+ hog submaxillary glycoprotein were set up with a C. perjringens preparation in the following three ways: (a) 1.21 pmoles of galactose-bound A+ glycoprotein alone, (6) the same amount of glycoprotein with the addition of 0.15 pmole of uniformly labeled radioactive [%]galactose, with a total count of 4 x lo5 cpm, (c) same as in b but with a further addition of 1.21 pmoles of nonradioactive galactose.
After 24 hours of incubation at 37" the incubation mixtures were heated in a boiling water bath for 1 min and alcohol was added to 80% (v/v).
The resulting supernatants were concentrated, aliquots were chromatographed beside suitable standards in Solvent 1 for 6 hours, and papers were well dried and cut into horizontal I-cm strips and counted in the Nuclear Chicago liquid scintillation counter using the toluene scintillation rnixture (31). The results obtained are shown in Table II and indicate quite clearly that all of the radioactivity is recovered at the RF of galactose, and less than 1% of the total radioactivity was found trailing in the oligosaccharide region.
The same was true if the chromatograms were stained with alkaline silver nitrate or Ehrlich reagent and the positively stained areas counted. Column a consisted of enzyme + substrnt,e + buffer, Column b was as Column a but with addition of radioactive [14C]galactose, and Column c was as Column b, with a further addition of 1.21 wmoles of galactose. The chromatograms were run in Solvent 1 on SS 589 paper for 6 hours. A+ oligosaccharide alditols, 1.13 prnoles of bound galactose plus fucose, were likewise incubated with the enzyme in the absence and presence of N- [1-14C]acetylgalactosarnine.
After 24 hours at 37" the incubation mixtures were treated as described above for radioactive galactose.
The 90% ethanol supernat,ants were concentrated and aliquots chromatographed to separate the free N-acetylgalactosarnine from the galactosyl-N-acetylgalactosaminide.
The appropriate areas of unstained strips were cut out and tested for ME + material by the intensity at 585 nm and radioactivity as in Section 3 above. The results are shown in Table III and again indicate that less than 1% of the total radioactivity was found trailing in t,he galactosyl-N-acetylgalactosaminide region.

Results with Other Glycoproteins
A nurnbcr of glycoproteins from miscellaneous sources were tested for their reactivity with the enzyme.
The glycoproteins were analyzed for their total hexosamine content after acid hydrolysis.
Incubation mixtures were set up to contain the same amount of total hexosamine, 5.5 prnoles.
After 24 hours at 3i", an aliquot was removed for the calorimetric deterrnination of total N-acetylhexosamine released, a second aliquot was treat,ed with ethanol (80%), and the supernatant concentrated and chromat,ographed in duplicate in Solvent 1 for 6 hours.
The two chromatograms were developed with alkaline silver nitrate and Ehrlich reagents (Table IV).
It has been previously reported that Clostridium welchii escretes enzymes capable of destroying A, I%, and O(H) activity (32). We have already purified and characterized the enzyme capable of destroying O(H) activity and shown it to be an (~(1 ---f 2).L-fucosidase (15). Turning our attention to the enzyme capable of destroying A activity, we had ventured to use the Morgan-Elson reaction to determine the free N-acetylgalactosamine released. .4s has been discussed previously (33), it is important to develop a suitable chemical assay for the purification of these enzymes rather than to depend on the sirnple loss of blood group activity.
For, as has been shown in the case of Clostridium fertium, the enzyme that destroys A activity turned out to be a deacetylase rather than an ol-N-acetylgalactosarninidase (34). Using the Morgan-Elson test to follow the hydrolysis of the A substance, it soon became apparent that it did not run parallel Quantitative estimation of the t,otal amount of ME+ material released on prolonged incubation indicated the production of excess color, calculated on the basis of the total galactosamine content of the glycoprotein.
The excess color could be attributed to a number of causes: (a) possible presence of an epimerase in the enzyme preparation capable of converting M-acetylgalactosamine to another, more reactive, N-acetylhexosamine, e.g. N-acetylglucosamine; (b) formation of N-glycolylmannosamine from the N-glycolylneuraminic acid present in t,he glycoprotein; and/or (c) the formation of an unsaturated oligosaccharide by a dehydrase action, as has been reported in the degradation of hyaluronic acid by C. welchii extracts (35-37). Use of an enzyme preparation free of sialidase or N-acetylneuraminic acid-aldolase, or the use of a substrate initially freed of sialic acid by treatment with pure sialidase still gave rise to excess ME + material.
The origin of the excess ME color could not therefore be attributed to the presence of sialic acid in the substrates.
Exploration of the other possibilities necessitated the separation of N-ace@1 galactosamine from other ME+ reacting materials.
This was readily achieved by paper chromatography. The successful separation of Component B (Table I) from free galactose and ME + reacting material by paper chromatography enabled us t,o purify it readily.
The product obtained was shown to be pure on rechromatography in Solvents 1 and 2 and on electrophoresis.
Qualitative and quantitative analysis identified the material as a disaccharide composed of equimolar amounts of galactose and N-acetylgalactosamine.
The react,ivity of t,he oligosaccharide in the ME test and the abolishment of that reactivity after reduction with sodium borohydride or oxidation with sodium hyl)oiodite identified the position of the hT-acct~~lgalactosa~nirlc at the reducing end of the disaccharide. This was further confirmed by (a) the inability of the galactose to react with galactose dehydrogenase, (until it was released by acid or enzymatic hydrolysis) and (b) the stability of the galactose to reduction with borohydride and oxidation with hypoiodite.
The nature of the anomeric linkage as /3 was established by its hydrolysis by @-galactosidase. Thus the structure of the ME+ reacting material is established as the disaccharide: 3-O-P-n-galactopyranosyl-2-acetamido%deoxy-ngalactose.
This structure is uniquely found at the reducing end of the many oligosaccharide chains in hog A, H, and In submaxillary glycoproteins (7, 13).
The oligosaccharide could, of course, arise also as a result of transgalactosylation activity (40) of the galactosidases known to be present in the C. perfringens extracts (15,41). The data provided readily exclude this possibility.
The isolation of this disaccharide from a number of glycoproteins that are both ilB0 blood group active and inactive, e.g. the glycoprotein from bovine colostrum and ColZacalitz mucin, indicates that the reducing end of many glycoproteins consists of this type of disaccharide.
Indeed this disaccharide has been isolated from acid hydrolysates of A, 13, H, and Lea (42, 43). Since it has been shown that the anotneric linkage between the reducing N-acetylgalactosamine and serine or threonine of the protein core is cr (44), we can presume that this enzyme is an a-n-N-acetylgalactosaminyl oligosaccharase. This is the first description of an oligosaccharase capable of breaking the protein-carbohydrate linkage in the undegraded glycoprotein to release a complete oligosaccharide. As such it should serve as a valuable tool in the elucidation of the structures of many glycoproteins of the blood group active type. The purification of the enzyme will be reported in a subsequent publication.
Acknowledgments-We are grateful to Dr. Jack Distler for the gas chromatographic analysis of galactose and the enzymatic hydrolysis of the oligosaccharide with P-galactosidase from bovine testes, and to Dr. Vincent C. Hascall for the determination of galactosamine on a Beckman amino acid analyzer.